There is a great warning in Against The Gods by Peter Bernstein describing games of chance played by the ancient Greeks and Romans with oblong dice. They attributed  special qualities to certain numbers that came up with greater frequency, not realizing that it was actually the larger surface area on the dice that accounted for these ‘special numbers.’

When we look at risk today – incorporating measures such as sensitivity analysis, VaR, and stress testing – are we similarly not missing out on the more fundamental phenomena and processes? 

In our highly connected global network of commerce events in disparate parts of the world can not only affect us but can do so rapidly. Events like the Covid-19 pandemic have shown us this both in the speed of its contagion as well as its impact on economies. Network analytics and visualizations are the way toward a more robust approach to risk management.

Having seen events such as the Russian default crisis and 9/11 unfold on a derivatives trading floor it became very clear that those who are unable to adapt and incorporate new information and methods would be highly penalized by the market. From the perspective of risk management a lot of the tried and true metrics which incorporate volatility, correlation and stress testing are at such unprecedented levels that it has become challenging to even contextualize the impact.

In 2020, it is imperative to have a  risk management framework that applies during ‘normal’ market conditions – as well as an additional set of measures for stressed or ‘extreme’ market conditions  complementing and completing the overall picture. Our traditional understanding of risk could be compared to the portion of the light spectrum that is visible to the naked eye. Infrared and X-ray vision are also a part of that spectrum, but we require additional tools to see them. By incorporating network metrics such as centrality (or influence), connectedness, and the powerful visualizations available from Network Science we can better understand and contextualize the risks and interconnectedness of financial assets.

Below is an example of a dashboard incorporating a visual representation of a portfolio of Exchange Traded Funds (ETF) that emphasizes interconnectedness of selected entities:

The tree-like representation of the entities (more specifically this is a Minimum Spanning Tree) allows a viewer to quickly focus on the important links between the ETFs by transforming the correlations into distances (entities linked closer represent higher correlations and those further apart represent lower correlations). 

Notice the close link between EAFE, Europe, France, Germany and Italy with the UK being further away. Interestingly, the branch including the Total Bond Index shows a connection to not only Corp Bonds and TIPS, but to JPY and Gold as well. 

We have identified each entity or node and whether the return on the asset is positive (green) or negative (red), with the size of the encompassing node or circle representing the size of the return. Similarly the links between the nodes are colour coded to represent positive or negative correlation. Notice on the S&P 500 branch the positive (green) connection to DJIA and less so to Tech, with a negative (red) correlation to the VIX.

Now that we have a good picture of the relevant interconnections we can take things a step further and begin to apply stress tests to both factors and correlations to see the resulting impact on our network. A risk practitioner can easily overlay ‘what-if’ scenarios by applying a shock say, to the S&P 500 as well as to correlations to see the resulting impact as indicated below:

Two interesting observations are that the network contracts and becomes more ‘tightly’ correlated – highlighting large negative returns in the equity based ETFs and large positive returns in the fixed income ETFs and VIX as a result. You are welcome to take a look at the dashboard here and play around with the stress testing capabilities.

As we can see, having a framework to incorporate contagion and interconnected risk is an important part of a holistic and evolving risk management program. This need not replace our traditional frameworks and practices; but can complement them with advanced tools for monitoring and visualizing various types of interconnected risk. At FNA we are focused on making advanced network analytics easily deployable to solve real, everyday business problems while helping to make the highly connected financial world more stable and efficient.

Mohsen Namazi ([email protected]) is a Managing Director of FNA based in North America.

More News

FNA Talks: Data Science in Economics and Finance for Decision Makers

Data Science in Economics and Finance for Decision Makers with ECB’s Per Nymand-Andersen.  In this latest episode of the FNA Talks podcast series, the ECB’s Per Nymand-Andersen joins FNA’s Suptech Lead, Adam Csabay to discuss one of the most awaited Fintech publications of 2021: Data Science in Economics and Finance for Decision Makers.  Per – […]

FNA contributes to Risk Book’s: Data Science in Economics and Finance for Decision Makers, Edited by Per Nymand-Andersen. 

FNA’ers Kimmo Soramäki, Ivana Ruffini, Mikhail Oet, Tuomas Takko and Adam Csabay contributed to the recent Risk Book: Data Science in Economics and Finance for Decision Makers, edited by the European Central Bank’s Per Nymand-Andersen.  The chapter FNA authored, Prudential Stress Testing in Financial Networks, provides a taxonomy of organizational problems facing firms operating in […]

Suptech Training & Events Update

During Winter 2021 we have been exceptionally busy with Suptech Training and Events for central banks and financial authorities. In February and March, we hosted the inaugural “Harnessing the Power of Suptech” Training Series – and very much enjoyed facilitating the engaging and thought-provoking discussions with 100+ delegates and guest speakers. In response to popular […]
Copyright FNA © 2021 | Privacy Policy